A computer-implemented method is provided for learning a model corresponding to a target function that changes in time series. The method includes acquiring a time-series parameter that is a time series of input parameters including parameter values expressing the target function. The method further includes propagating propagation values, which are obtained by weighting parameters values at time points before one time point according to passage of the time points, to nodes in the model associated with the parameter values at the one time point. The method also includes calculating a node value of each node using each propagation value propagated to each node. The method additionally includes updating a weight parameter used for calculating the propagation values propagated to each node, using a difference between the target function at the one time point and a prediction function obtained by making a prediction from the node values of the nodes.
Legal claims defining the scope of protection, as filed with the USPTO.
1. A computer-implemented method for learning a model corresponding to a geographic-based target function, comprising: acquiring a time-series parameter that is a time series of input parameters that include a plurality of parameter values expressing the geographic-based target function; propagating each of a plurality of propagation values, which are obtained by weighting each of the plurality of parameters values at a plurality of time points before one time point according to a passage of the plurality of time points, to a plurality of nodes in the model associated with the plurality of parameter values at the one time point; calculating a node value of each of the plurality of nodes using a respective one of the plurality of propagation values propagated to each of the plurality of nodes; and updating a weight parameter used for calculating the plurality of propagation values propagated to the plurality of nodes, using a difference between the geographic-based target function at the one time point and a prediction function obtained by making a prediction from node values of the plurality of nodes.
2. The computer-implemented method of claim 1 , further comprising for the one time point, acquiring an output value of the geographic-based target function corresponding to each input value in a first plurality of input values for the geographic-based target function, wherein the updating step includes updating the weight parameter using a difference between the output value of the geographic-based target function corresponding to the each input value in the first plurality of input values and an output value of the prediction function, at the one time point.
3. The computer-implemented method of claim 2 , wherein each of the plurality of nodes corresponds to an input value in a second plurality of input values in a defined region of the geographic-based target function.
4. The computer-implemented method of claim 3 , wherein at least a subset of the first plurality of input values and the second plurality of input values are different.
5. The computer-implemented method of claim 4 , wherein the updating step includes calculating the output value of the prediction function corresponding to the each input value in the first plurality of input values, from each node value of the plurality of nodes corresponding to the input values in the second plurality of input values.
6. The computer-implemented method of claim 5 , wherein the acquiring step includes calculating the plurality of parameter values at the one time point based on the node values of the plurality of nodes and a difference between the output value of the prediction function and the output value of the geographic-based target function corresponding to the each input value in the first plurality of input values.
7. The computer-implemented method of claim 4 , wherein at least a subset of the first plurality of input values are different at each time point.
8. The computer-implemented method of claim 2 , wherein the updating step includes performing the updating of the weight parameter corresponding to the one time point on a condition that a number of input values in the first plurality of input values exceeds a threshold value at the one time point.
9. The computer-implemented method of claim 2 , wherein the updating step includes changing a learning rate of the weight parameter according to a number of input values in the first plurality of input values.
10. The computer-implemented method of claim 1 , wherein the geographic-based target function has a position in a one-dimensional or multi-dimensional space input thereto, and outputs a value relating to the position.
11. A computer-implemented method of using a model that has been learned corresponding to a geographic-based target function, the method comprising: acquiring a time-series parameter that is a time series of input parameters that include a plurality of parameter values expressing the geographic-based target function; propagating each of a plurality of propagation values, which are obtained by weighting each of the plurality of parameters values at a plurality of time points before one time point according to a passage of the plurality of time points, to a plurality of nodes in the model associated with the plurality of parameter values at the one time point; calculating a node value of each of the plurality of nodes using a respective one of the plurality of propagation values propagated to each of the plurality of nodes; and calculating a prediction function that is a prediction of the geographic-based target function at the one time point from node values of the plurality of nodes.
12. The computer-implemented method of claim 11 , further comprising calculating an output value of the prediction function corresponding to each input value in a first plurality of input values.
13. The computer-implemented method of claim 12 , wherein each of the plurality of nodes corresponds to an input value in a second plurality of input values in a defined region of the geographic-based target function.
14. The computer-implemented method of claim 13 , wherein at least some of the first plurality of input values and the second plurality of input values are different.
15. The computer-implemented method of claim 11 , further comprising updating a weight parameter used for calculating the respective one of the plurality of propagation values propagated to each of the plurality of nodes, using a difference between the prediction function and the geographic-based target function at the one time point.
16. A non-transitory computer readable storage medium having instructions embodied therewith, the instructions executable by a processor or programmable circuitry to cause the processor or programmable circuitry to perform a method, the method comprising: acquiring a time-series parameter that is a time series of input parameters including a plurality of parameter values expressing a geographic-based target function; propagating each of a plurality of propagation values, which are obtained by weighting each of a plurality of parameters values at a plurality of time points before one time point according to passage of the time points, to a plurality of nodes in a model corresponding to the geographic-based target function associated with the plurality of parameter values at the one time point; calculating a node value of each of the plurality of nodes using a respective one of the plurality of propagation values propagated to each of the plurality of nodes; and updating a weight parameter used for calculating the propagation values propagated to each of the plurality of nodes, using a difference between the geographic-based target function at the one time point and a prediction function obtained by making a prediction from node values of the plurality of nodes.
17. The non-transitory computer readable storage medium of claim 16 , wherein the method further comprises, for the one time point, acquiring an output value of the geographic-based target function corresponding to each input value in a first plurality of input values for the geographic-based target function, and wherein the updating step includes updating the weight parameter using a difference between the output value of the geographic-based target function corresponding to the each input value in the first plurality of input values and an output value of the prediction function, at the one time point.
18. The non-transitory computer readable storage medium of claim 17 , wherein each of the plurality of nodes corresponds to an input value in a second plurality of input values in a defined region of the geographic-based target function.
19. The non-transitory computer readable storage medium of claim 18 , wherein at least a subset of the first plurality of input values and the second plurality of input values are different.
20. An apparatus comprising: a processor or programmable circuitry operable to execute instructions, a non-transitory computer readable storage medium having the instructions embodied therewith, the instructions executable by the processor or programmable circuitry to cause the processor or programmable circuitry to perform a method; and wherein the method includes: acquiring a time-series parameter that is a time series of input parameters including a plurality of parameter values expressing a geographic-based target function; propagating each of a plurality of propagation values, which are obtained by weighting each of a plurality of parameters values at a plurality of time points before one time point according to passage of the time points, to a plurality of nodes in a model corresponding to the geographic-based target function associated with the plurality of parameter values at the one time point; calculating a node value of each of the plurality of nodes using a respective one of the plurality of propagation values propagated to each of the plurality of nodes; and updating a weight parameter used for calculating the propagation values propagated to each of the plurality of nodes, using a difference between the geographic-based target function at the one time point and a prediction function obtained by making a prediction from node values of the plurality of nodes.
21. A non-transitory computer readable storage medium having instructions embodied therewith, the instructions executable by a processor or programmable circuitry to cause the processor or programmable circuitry to perform a method, the method comprising: acquiring a time-series parameter that is a time series of input parameters including a plurality of parameter values expressing a geographic-based target function; propagating each of a plurality of propagation values, which are obtained by weighting each of a plurality of parameters values at a plurality of time points before one time point according to passage of the time points, to a plurality of nodes in a model corresponding to the geographic-based target function associated with the plurality of parameter values at the one time point; calculating a node value of each of the plurality of nodes using a respective one of the plurality of propagation values propagated to each node; and calculating a prediction function that is a prediction of the geographic-based target function at the one time point from the node values of the plurality of nodes.
22. The non-transitory computer readable storage medium of claim 21 , wherein the method further comprises calculating an output value of the prediction function corresponding to each input value in a first plurality of input values.
23. The non-transitory computer readable storage medium of claim 22 , wherein each of the plurality of nodes corresponds to an input value in a second plurality of input values in a defined region of the geographic-based target function.
24. The non-transitory computer readable storage medium of claim 21 , wherein the method further comprises updating a weight parameter used for calculating the respective one of the plurality of propagation values propagated to each of the plurality of nodes, using a difference between the prediction function and the geographic-based target function at the one time point.
25. An apparatus comprising: a processor or programmable circuitry operable to execute instructions, a non-transitory computer readable storage medium having the instructions embodied therewith, the instructions executable by the processor or programmable circuitry to cause the processor or programmable circuitry to perform a method; and wherein the method includes: acquiring a time-series parameter that is a time series of input parameters including a plurality of parameter values expressing a geographic-based target function; propagating each of a plurality of propagation values, which are obtained by weighting each of a plurality of parameters values at a plurality of time points before one time point according to passage of the time points, to a plurality of nodes in a model corresponding to the geographic-based target function associated with the plurality of parameter values at the one time point; calculating a node value of each of the plurality of nodes using a respective one of the plurality of propagation values propagated to each node; and calculating a prediction function that is a prediction of the geographic-based target function at the one time point from the node values of the plurality of nodes.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
April 28, 2017
September 22, 2020
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.